pkgsrc-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Problem with stunnel in pkgsrc



On 8/12/24 10:54 AM, Greg Troxel wrote:
Jason Mitchell <jmitchel%bigjar.com@localhost> writes:

     I've been dealing with a problem where stunnel segfaults on every
connection where OCSP stapling is required. The stunnel folks just got
back to me and there's an easy fix -- add stack = 131072 and the
problem goes away. There going to fix it in the next release. Should
we put a warning message on the stunnel package in pkgsrc advising
people to add this line to their stunnel.conf file?
It would make sense to add this to the example config file that the
package does or should install.

Or, to see what their fix is and apply it as a patch.  I wonder if their
approach is to increase the limit, or to be less consuming of resources.

ulimit -s shows

stack size                  (kbytes, -s) 4096

on a system where I have no memory of messing with stack size.
That's 4 MB, which seems like quite a lot of stack.

I wonder if they really mean 128 MB of stack?

Greg,

    Here's the message I sent to netbsd-users with the message from the stunnel developer at the bottom. Maybe it's an internal stack?


https://mail-index.netbsd.org/netbsd-users/2024/08/12/msg031635.html


    I tracked down the problem to stunnel's resolver.c (called from ocsp.c), but stopped when I saw that the stunnel folks had identified the problem and had a fix in the works as well as a workaround. If we're including stack = 131072 in the stunnel.conf file then we probably should add a comment so that people know why it's there. For example:


; The following line is necessary to work around a bug in stunnel related to OCSP stapling. This bug should be fixed in stunnel 5.73.

stack = 131072


    Should I file a PR?

Thanks,

Jason M.



Home | Main Index | Thread Index | Old Index